This invention relates to the field of digital imaging analysis and more specifically to digital rock physics and methods for obtaining improved values for rock properties derived from digital images.
Acquiring, developing and managing hydrocarbon reservoirs involve many decisions that are sensitive to the quality of information about the physical properties of the reservoir rock. For instance, accurate assessments of porosity, absolute permeability, relative permeability, capillary pressure, electrical resistivity, and elastic properties of rock under investigation are each of interest and value to reservoir engineers in applications including well planning, completion design, and reservoir estimates. Detailed, specific information on rock structure and properties can be useful on its own or can facilitate leveraging the greatest value out of logging and seismic data to which it provides important context. Some of the traditional techniques to acquire information about the reservoir rock have determined properties by measuring overall effect, e.g., permeability might be established with a permeameter forcing a fluid through a rock sample and recording the resulting fluid flux and pressure drops. However, such attempts to acquire information may be limited by the shape and size of the sample and are often otherwise not well suited to providing quality information in a timely manner.
Digital rock physics is an important tool for facilitating a better, quicker, and more efficient insight into rock structure and properties of interest. Such techniques produce values of important rock properties through analysis of three-dimensional (3D) images, also referred to as volumes or digital objects, representing the natural rock samples. Original grey-scale image acquisition uses scanning operations such as X-Ray computed tomography scans (CT scans) or with a focused ion beam scanning electron microscope (FIB-SEM scanning). Such techniques integrate a succession of cross-sectional scans into a whole 3D image. This original grey-scale image can be processed with segmentation techniques through subdividing the volume into discrete sub volumes, individual voxels at the most elementary level, and processing these to produce segmented volumes with each voxel allocated to either pore or a variety of solid material phases. As used herein, “solid material phases” means the mixture of grains of various minerals, cementation components, and all else that is not pore space as imaged by the scan.
A variety of segmentation methods are known to those skilled in the art of digital rock physics. These segmentation methods include, for example, those shown by Toelke, J., et al. (2010), “Computer simulations of fluid flow in sediment: From images to permeability,” The Leading Edge (January 2010), 68-74, (hereinafter, the “Toelke (2010)” publication), and U.S. Pat. No. 8,081,802 B2 to Dvorkin et al. (hereinafter, the '802 patent).
Values for important properties can be estimated, modeled, or simulated with the resulting segmented volumes. See, e.g., Dvorkin, J., et al. (2011), “Relevance of computation rock physics,” Geophysics, 76(5), E141-E153 (hereinafter, the “Dvorkin 2011” publication) and the '802 patent.
Circumstances are sometimes encountered, however, in which the value of one or more properties derived only from segmented volumes is unreliable. One such circumstance capable of producing unreliable property values occurs when important structure is below the resolution capabilities. These limitations may commonly be the result of the need to work with manageably sized data sets for computations and simulations and in this case derive from the scanning resolution in combination with the field-of-view required to address a representative sample. Alternatively, in some cases structural features contributing to properties of interest may be too small to accurately capture in scanned images as a limitation of the scanning equipment itself, such as sub-resolution or under-resolved features. Either way, there are instances in which an important structure is not directly captured, e.g., very thin conduits connecting the pores in some samples may not accurately resolve even at the highest magnifications practically possible. This may result in a segmented volume which does not adequately represent the actual structure of the rock.
In such cases, usual analytical techniques deriving values from such volume can yield unreliable values, e.g., an absolute permeability k for the rock that is unrealistically small or even at zero where the actual effective characteristics for that rock are quite different, e.g., production data proves otherwise. Values for elastic properties, relative permeability, capillary pressure, electrical resistivity, and other rock properties may be similarly affected. Greater accuracy for such values is important for decisions critical to determining what reservoir zones are of commercial interest and how to develop a field.
These resolution issues have become a well-recognized challenge. Dvorkin (2011) discusses the problem (e.g., E144), as does Toelke (2010) in the discussion of “unresolved pores” (p. 70) thereof, and Knackstedt, M. A., et al. (2004), “Digital core laboratory: Properties of reservoir core derived from 3D images,” SPE Asian Pacific Conference on Integrated Modeling for Asset Management. SPE 87009, discusses the problem. It has been proposed to address such sub-resolution and/or under-resolved features by adjusting the mathematical models that are used to derive property values from the segmented volumes. See, e.g., Toelke (2010), and the published European Patent Application publication no. EP 2090907A1 for a Method for determining the properties of hydrocarbon reservoirs from geophysical data, for discussions of adjusting the mathematical models and/or changing values ascribed to phases. These adjustments are limited and can fail to produce digital volumes with structural features representative of the rock under investigation.
The use of filters with the grey-scale image and a seeding/region growing segmentation approach such as disclosed in the '802 Patent can assist effective image processing by removing various anomalies and smoothing data. However, images so addressed still fail to capture connections between the pore spaces that account for properties present but which result from structure below the full resolution of the scanner.
A watershed transform had also been applied to further manipulate digital volumes. See, generally, Vincent, L., et al. (1991), “Watershed in digital spaces: An efficient algorithm based on immersion simulations,” IEEE Transactions on Pattern Analysis and Machine Intelligence. 13(6), 583-598 (hereinafter, the “Vincent (1991)” publication) and Faessel, M., et al., (2009), “Touching grain kernels separation by gap-filling,” Image Anal Stereol. 29. 195-203., which introduce application of the watershed transform and application to the inverse of the distance function and to gap-filling techniques. See, e.g., Sakellariou, A., et al. (2007), “Developing a virtual materials laboratory,” Materials Today, 10(12), 44-51, which discusses using watershed to decompose an object. Quintal, B., et al. (2011), “Integrated numerical and laboratory rock physics applied to seismic characterization of reservoir rocks,” The Leading Edge, (December 2011), 1360-1376 (hereinafter, the “Quintal (2011)” publication), relates to localizing the grain contacts below the data resolution, a “grain-contact reconstruction” method. However, the Quintal (2011) application of watershed produces unrealistic results due to “unresolved microcracks and other microstructures, which cannot be detected with the grain contact reconstruction technique” and then shows alteration of the mathematical model to compensate.
As in other prior approaches, such compensation can facilitate producing an improved value, but does not correct the structure featured in the segmented volume to best represent the rock sample.
Thus, the present investigators have recognized that there is a need for developing a method for adjusting the segmented volume to effectively amend the structure featured therein to represent the rock sample more realistically and thereby produce more reliable property values.